Dev Fest
Jupyter Notebook HTML CSS Python JavaScript
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
.ipynb_checkpoints
app
data
photos_deb
.gitignore
HilBer.jpg
Notebook to call the API Emotion and Face.ipynb
Photo_API_MSoxford.ipynb
Readme.md
Untitled.ipynb
alchemyapi.py
alchemyapi.pyc
api_key.txt
bernie_sshot.png
combined_results.csv
combined_results_withGender.csv
combined_results_withGender.json
data.json
emotion_results.csv
emotions.ipynb
face_API_MSoxford.py
hillary_sshot.png
image_results.csv
matching text frame.csv
matching text frame.json
mongodb.ipynb
text.ipynb
text_collection.json
transcript.ipynb
video-api.ipynb

Readme.md

Debate in Emotion

This project is the result of the #devfest16 hackathon at Columbia University, which took place on February 5th, 2016.

The goal of this project is to analyse the variation in presidential candidates' emotions during debates. For that matter, we took as a sample the debate between Hillary Clinton and Bernie Sanders, which took place on February 4th, 2016.

Samples were taken from the following video:

IMAGE ALT TEXT

Sample screenshots

The graph below shows the emotion levels of both candidates over time, during the two-hour long presidential debate on February 4th. Emotions range from anger to happiness.

alt hillary

Hovering over specific points show the estimated emotion levels on a video capture at a specific time. Additionally, emotions can be turned off and on in order to clarify the plots.

alt bernie_sanders

Usage

Just open app/index.html in a browser. Safari and Firefox are recommended. There are known issues with Google Chrome, which we hope to fix soon.

APIs

This project relies heavily on the following APIs:

  • Microsoft® Project Oxford Video API
  • Microsoft® Project Oxford Emotion API
  • Alchemy® Entity Extraction API

Technologies

  • All data processing was done in Python 3 with MongoDB as the database, connecting through pyMongo.
  • Text and grep-hacking was done using Sublime Text 3 and Atom.
  • The front-end has been implemented using jQuery, d3.js and nvd3.
  • The video file was split sequentially using Matlab.

The Team

We're MS in Data Science students at Columbia University. We go by the following names:

  • Carlos Espino (Mexico)
  • Amirhos Imani (Iran)
  • Xavier González (Argentina)
  • Diego Llarrull (Argentina)